Visualization onboarding supports users in reading, interpreting, and extracting information from visual data representations. General-purpose onboarding tools and libraries are applicable for explaining a wide range of graphical user interfaces but cannot handle specific visualization requirements. This paper describes a first step towards developing an onboarding library called VisAhoi, which is easy to integrate, extend, semi-automate, reuse, and customize . VisAhoi supports the creation of onboarding elements for different visualization types and datasets. We demonstrate how to extract and describe onboarding instructions using three well-known high-level descriptive visualization grammars — Vega-Lite, Plotly.js, and ECharts. We show the applicability of our library by performing two usage scenarios that describe the integration of VisAhoi into a VA tool for the analysis of high-throughput screening (HTS) data and, second, into a Flourish template to provide an authoring tool for data journalists for a treemap visualization. We provide a supplementary website ( https://datavisyn.github.io/visAhoi/ ) that demonstrates the applicability of VisAhoi to various visualizations, including a bar chart, a horizon graph, a change matrix/heatmap, a scatterplot, and a treemap visualization.

Understanding user behavior patterns and visual analysis strategies is a long-standing challenge. Existing approaches rely largely on time-consuming manual processes such as interviews and the analysis of observational data. While it is technically possible to capture a history of user interactions and application states, it remains difficult to extract and describe analysis strategies based on interaction provenance. In this article, we propose a novel visual approach to the meta-analysis of interaction provenance. We capture single and multiple user sessions as graphs of high-dimensional application states. Our meta-analysis is based on two different types of two-dimensional embeddings of these high-dimensional states: layouts based on (i) topology and (ii) attribute similarity. We applied these visualization approaches to synthetic and real user provenance data captured in two user studies. From our visualizations, we were able to extract patterns for data types and analytical reasoning strategies.

Abstract Dashboards are used ubiquitously to gain and present insights into data by means of interactive visualizations. To bridge the gap between non‐expert dashboard users and potentially complex datasets and/or visualizations, a variety of onboarding strategies are employed, including videos, narration, and interactive tutorials. We propose a process model for dashboard onboarding that formalizes and unifies such diverse onboarding strategies. Our model introduces the onboarding loop alongside the dashboard usage loop. Unpacking the onboarding loop reveals how each onboarding strategy combines selected building blocks of the dashboard with an onboarding narrative. Specific means are applied to this narration sequence for onboarding, which results in onboarding artifacts that are presented to the user via an interface. We concretize these concepts by showing how our process model can be used to describe a selection of real‐world onboarding examples. Finally, we discuss how our model can serve as an actionable blueprint for developing new onboarding systems.

Classifiers are among the most widely used supervised machine learning algorithms. Many classification models exist, and choosing the right one for a given task is difficult. During model selection and debugging, data scientists need to assess classifiers' performances, evaluate their learning behavior over time, and compare different models. Typically, this analysis is based on single-number performance measures such as accuracy. A more detailed evaluation of classifiers is possible by inspecting class errors. The confusion matrix is an established way for visualizing these class errors, but it was not designed with temporal or comparative analysis in mind. More generally, established performance analysis systems do not allow a combined temporal and comparative analysis of class-level information. To address this issue, we propose ConfusionFlow, an interactive, comparative visualization tool that combines the benefits of class confusion matrices with the visualization of performance characteristics over time. ConfusionFlow is model-agnostic and can be used to compare performances for different model types, model architectures, and/or training and test datasets. We demonstrate the usefulness of ConfusionFlow in a case study on instance selection strategies in active learning. We further assess the scalability of ConfusionFlow and present a use case in the context of neural network pruning.

Comprehending and exploring large and complex data is becoming increasingly important for a diverse population of users in a wide range of application domains. Visualization has proven to be well-suited in supporting this endeavor by tapping into the power of human visual perception. However, non-experts in the field of visual data analysis often have problems with correctly reading and interpreting information from visualization idioms that are new to them. To support novices in learning how to use new digital technologies, the concept of onboarding has been successfully applied in other fields and first approaches also exist in the visualization domain. However, empirical evidence on the effectiveness of such approaches is scarce. Therefore, we conducted three studies with Amazon Mechanical Turk (MTurk) workers and students investigating visualization onboarding at different levels: (1) Firstly, we explored the effect of visualization onboarding, using an interactive step-by-step guide, on user performance for four increasingly complex visualization techniques with time-oriented data: a bar chart, a horizon graph, a change matrix, and a parallel coordinates plot. We performed a between-subject experiment with 596 participants in total. The results showed that there are no significant differences between the answer correctness of the questions with and without onboarding. Particularly, participants commented that for highly familiar visualization types no onboarding is needed. However, for the most unfamiliar visualization type — the parallel coordinates plot — performance improvement can be observed with onboarding. (2) Thus, we performed a second study with MTurk workers and the parallel coordinates plot to assess if there is a difference in user performances on different visualization onboarding types: step-by-step, scrollytelling tutorial, and video tutorial. The study revealed that the video tutorial was ranked as the most positive on average, based on a sentiment analysis, followed by the scrollytelling tutorial and the interactive step-by-step guide. (3) As videos are a traditional method to support users, we decided to use the scrollytelling approach as a less prevalent way and explore it in more detail. Therefore, for our third study, we gathered data towards users’ experience in using the in-situ scrollytelling for the VA tool Netflower. The results of the evaluation with students showed that they preferred scrollytelling over the tutorial integrated in the Netflower landing page. Moreover, for all three studies we explored the effect of task difficulty. In summary, the in-situ scrollytelling approach works well for integrating onboarding in a visualization tool. Additionally, a video tutorial can help to introduce interaction techniques of visualization.

Comprehending and exploring large and complex data is becoming increasingly important for a diverse population of users in a wide range of application domains. Visualization has proven to be well-suited in supporting this endeavor by tapping into the power of human visual perception. However, non-experts in the field of visual analysis often have difficulties in correctly reading and interpreting information from novel visualization idioms. Visualization onboarding can support novices in learning how to use new digital technologies. Therefore, we developed an interactive step-by-step guide and applied the method to four visualization techniques—a bar chart, a horizon graph, a change matrix, and a parallel coordinates plot. Results using Amazon Mechanical Turk workers show that there is a need for onboarding, especially for more complex visualization techniques. We further investigated the perception and rating of a scrollytelling and a video tutorial for the most unfamiliar visualization—the parallel coordinates plot. A comparison between the three onboarding methods indicates that participants appreciated the easy-to-understand examples, the precise wording of the onboarding messages in a step-by-step manner, and the introduction of interaction concepts by highlighting the most relevant information over all onboarding methods. The video tutorial supported the introduction of unknown interaction techniques best.

In problem-solving, a path towards a solutions can be viewed as a sequence of decisions. The decisions, made by humans or computers, describe a trajectory through a high-dimensional representation space of the problem. By means of dimensionality reduction, these trajectories can be visualized in lower-dimensional space. Such embedded trajectories have previously been applied to a wide variety of data, but analysis has focused almost exclusively on the self-similarity of single trajectories. In contrast, we describe patterns emerging from drawing many trajectories—for different initial conditions, end states, and solution strategies—in the same embedding space. We argue that general statements about the problem-solving tasks and solving strategies can be made by interpreting these patterns. We explore and characterize such patterns in trajectories resulting from human and machine-made decisions in a variety of application domains: logic puzzles (Rubik’s cube), strategy games (chess), and optimization problems (neural network training). We also discuss the importance of suitably chosen representation spaces and similarity metrics for the embedding.

SUMMARY: Ordino is a web-based analysis tool for cancer genomics that allows users to flexibly rank, filter and explore genes, cell lines and tissue samples based on pre-loaded data, including The Cancer Genome Atlas, the Cancer Cell Line Encyclopedia and manually uploaded information. Interactive tabular data visualization that facilitates the user-driven prioritization process forms a core component of Ordino. Detail views of selected items complement the exploration. Findings can be stored, shared and reproduced via the integrated session management. AVAILABILITY AND IMPLEMENTATION: Ordino is publicly available at https://ordino.caleydoapp.org. The source code is released at https://github.com/Caleydo/ordino under the Mozilla Public License 2.0. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Storing analytical provenance generates a knowledge base with a large potential for recalling previous results and guiding users in future analyses. However, without extensive manual creation of meta information and annotations by the users, search and retrieval of analysis states can become tedious. We present KnowledgePearls, a solution for efficient retrieval of analysis states that are structured as provenance graphs containing automatically recorded user interactions and visualizations. As a core component, we describe a visual interface for querying and exploring analysis states based on their similarity to a partial definition of a requested analysis state. Depending on the use case, this definition may be provided explicitly by the user by formulating a search query or inferred from given reference states. We explain our approach using the example of efficient retrieval of demographic analyses by Hans Rosling and discuss our implementation for a fast look-up of previous states. Our approach is independent of the underlying visualization framework. We discuss the applicability for visualizations which are based on the declarative grammar Vega and we use a Vega-based implementation of Gapminder as guiding example. We additionally present a biomedical case study to illustrate how KnowledgePearls facilitates the exploration process by recalling states from earlier analyses.

Multivariate, tabular data is one of the most common data structures used in many different domains. Over time, tables can undergo changes in both structure and content, which results in multiple versions of the same table. A challenging task when working with such derived tables is to understand what exactly has changed between versions in terms of additions/deletions, reorder, merge/split, and content changes. For textual data, a variety of commonplace "diff" tools exist that support the task of investigating changes between revisions of a text. Although there are some comparison tools which assist users in inspecting differences between multiple table instances, the resulting visualizations are often difficult to interpret or do not scale to large tables with thousands of rows and columns. To address these challenges, we developed TACO, an interactive comparison tool that visualizes the differences between multiple tables at various levels of detail. With TACO we show (1) the aggregated differences between multiple table versions over time, (2) the aggregated changes between two selected table versions, and (3) detailed changes between the selected tables. To demonstrate the effectiveness of our approach, we show its application by means of two usage scenarios.

A major challenge in data-driven biomedical research lies in the collection and representation of data provenance information to ensure that findings are reproducibile. In order to communicate and reproduce multi-step analysis workflows executed on datasets that contain data for dozens or hundreds of samples, it is crucial to be able to visualize the provenance graph at different levels of aggregation. Most existing approaches are based on node-link diagrams, which do not scale to the complexity of typical data provenance graphs. In our proposed approach, we reduce the complexity of the graph using hierarchical and motif-based aggregation. Based on user action and graph attributes, a modular degree-of-interest (DoI) function is applied to expand parts of the graph that are relevant to the user. This interest-driven adaptive approach to provenance visualization allows users to review and communicate complex multi-step analyses, which can be based on hundreds of files that are processed by numerous workflows. We have integrated our approach into an analysis platform that captures extensive data provenance information, and demonstrate its effectiveness by means of a biomedical usage scenario.

Multi-attribute time-series data plays a vital role in many different domains, such as economics, sensor networks, and biology. An important task when making sense of such data is to provide users with an overview to identify items that show an interesting development over time, including both absolute and relative changes in multiple attributes simultaneously. However, this is not well supported by existing visualization techniques. To address this issue, we present ThermalPlot, a visualization technique that summarizes combinations of multiple attributes over time using an items position, the most salient visual variable. More precisely, the x-position in the ThermalPlot is based on a user-defined degree-of-interest (DoI) function that combines multiple attributes over time. The y-position is determined by the relative change in the DoI value ( ∆DoI) within a user-specified time window. Animating this mapping via a moving time window gives rise to circular movements of items over time-as in thermal systems. To help the user to identify important items that match user-defined temporal patterns and to increase the technique's scalability, we adapt the level of detail of the items' representation based on the DoI value. Furthermore, we present an interactive exploration environment for multi-attribute time-series data that ties together a carefully chosen set of visualizations, designed to support analysts in interacting with the ThermalPlot technique. We demonstrate the effectiveness of our technique by means of two usage scenarios that address the visual analysis of economic development data and of stock market data.

With the rise of virtualization and cloud-based networks of various scales and degrees of complexity, new approaches to managing such infrastructures are required. In these networks, relationships among components can be of arbitrary cardinality (1:1, 1:n, n:m), making it challenging for administrators to investigate which components influence others. In this paper we present CloudGazer, a scalable visualization system that allows users to monitor and optimize cloud-based networks effectively to reduce energy consumption and to increase the quality of service. Instead of visualizing the overall network, we split the graph into semantic perspectives that provide a much simpler view of the network. CloudGazer is a multiple coordinated view system that visualizes either static or live status information about the components of a perspective while reintroducing lost inter-perspective relationships on demand using dynamically created inlays. We demonstrate the effectiveness of CloudGazer in two usage scenarios: The first is based on a real-world network of our domain partners where static performance parameters are used to find an optimal design. In the second scenario we use the VAST 2013 Challenge dataset to demonstrate how the system can be employed with live streaming data.

Since a couple of years tablets are gaining ever increasing markets share, so that desktop PCs are slowly losing their dominant role. People with disabilities are also attracted by tablets but they are sometimes challenged by the multi-touch technology itself and the natural interfaces of the provided applications.